Employees' reactions to problematic events: a circumplex structure of [PDF]

Journal of Organizational Behavior J. Organiz. Behav. 20, 309±321 (1999)

Employees' reactions to problematic events: a circumplex structure of ®ve categories of responses, and the role of job satisfaction MARIEÈT HAGEDOORN* , NICO W. VAN YPEREN, EVERT VAN DE VLIERT AND BRAM P. BUUNK Department of Psychology, University of Groningen, Grote Kruisstraat 2/1, 9712 TS Groningen, The Netherlands

Summary

A re®ned exit, voice, loyalty, and neglect (EVLN) typology (Farrell, 1983) was examined. It was argued that the category of voice responses, that is, attempts to improve the situation, should be divided into two forms: considerate voice and aggressive voice. Considerate voice consists of attempts to solve the problem taking into account one's own concerns as well as those of the organization, and aggressive voice consists of e€orts to win, without consideration for the concerns of the organization. In line with Farrell (1983), all categories of behavioral responses were assumed to di€er on two dimensions, namely, destructive±constructive and active±passive. It was assumed that the responses can be ordered in a circumplex structure. Factor analyses of data from 233 teachers and maternity nurses demonstrated the empirical separability of the ®ve categories of behavioral responses and supported the two-dimensional structure of the model. Job satisfaction, especially satisfaction with supervision, seemed to promote considerate voice and loyalty (that was relabeled as patience), and suppress exit, aggressive voice, and neglect. Furthermore, the correlations between job satisfaction and the ®ve responses provided additional support for the circumplex structure of the re®ned model. The pattern of responses runs in the following order: considerate voice, aggressive voice, exit, neglect, patience. Copyright # 1999 John Wiley & Sons, Ltd.

Introduction Every now and then, employees are confronted with problems at work, such as di€erences of opinion with a superior, frustrations about assignments or tasks, and irritations because of the behavior of co-workers. The way in which employees deal with these problematic events that produce negative feelings may have an important impact on their relationship with the organization and eventually the e€ectiveness of the organization. For instance, reporting a problem to a supervisor may have more bene®cial e€ects than avoiding the problem by staying away from work because reporting the problem may reduce distress and may alert the organization that some procedure or practice is not working (Sheppard, Lewicki and Minton, 1992). * Correspondence to: M. Hagedoorn, Department of Psychology, University of Groningen, Grote Kruisstraat 2/1, 9712 TS Gronningen, The Netherlands.

CCC 0894±3796/99/030309±13$17.50 Copyright # 1999 John Wiley & Sons, Ltd.

Accepted 26 March 1997

310

M. HAGEDOORN ET AL.

The literature from several ®elds within organizational psychology provides information on the di€erent ways in which employees may respond to problematic events. A very relevant line of research concerns studies on reactions to dissatisfaction and the exit, voice, loyalty, and neglect (EVLN) typology (Farrell, 1983; Fischer and Locke, 1992; Hirschman, 1970; Hulin, 1991; Leck and Saunders, 1992; Rusbult, Farrell, Rogers and Mainous, 1988; Withey and Cooper, 1989). Other important contributions come from studies on stress and coping in general (for a review, see Kahn and Byosiere, 1992) and con¯ict management in particular (e.g. Blake and Mouton, 1964; Rubin, Pruitt and Kim, 1994). However, the studies within these di€erent paradigms have been conducted in relative isolation from each other. Therefore, in the current study, the primary purpose was to gain greater insight into the di€erent reactions to problematic events at work by integrating ideas from the above-mentioned ®elds of organizational psychology. To this end, on the basis of the literature on con¯ict management, we re®ned and tested the EVLN typology. Farrell (1983) proposed a typology of four possible categories of behavioral responses to dissatisfaction. This typology, which is based on the work of Hirschman (1970), consists of exit (E), voice (V), loyalty (L), and neglect (N). Exit includes leaving the organization, thinking about quitting and looking for another job. Voice is de®ned as trying to change the situation, for example, by discussing the problem with the supervisor, by suggesting solutions, by seeking help from an outside agency, or whistle-blowing. Loyalty means patiently waiting for conditions to improve, trusting the organization to do the right thing. Behaviors such as chronic lateness or absenteeism and using company time for personal business are labeled neglect. Farrell (1983; Rusbult et al., 1988) described these so-called EVLN categories along two dimensions. The ®rst is the destructive±constructive dimension, and the second is the active± passive dimension. Both voice and loyalty are categorized as constructive behaviors because they are aimed at maintaining or reviving the relationship with the organization. In contrast, exit and neglect are viewed as destructive to this relationship. With respect to the second dimension, active is de®ned as dealing directly with the problematic situation. Exit and voice are therefore considered as active behaviors (`shape up or ship out'), and neglect and loyalty are seen as passive behaviors.1 Recently, Leck and Saunders (1992) argued that it would be better to relabel loyalty as patience because the term loyalty is also used to describe an attitude, while patience is more appropriate to describe a behavior; patience describes the act of waiting optimistically better than loyalty. Therefore, in the remainder of this article, the term patience will be used instead of loyalty. The EVLN typology has been supported in several studies (Farrell, 1983; Farrell and Rusbult, 1992; Leck and Saunders, 1992; Rusbult, Farrell, Rogers and Mainous, 1988; Rusbult and Lowery, 1985; Withey and Cooper, 1989), but there are two issues that need to be addressed. Firstly, according to the EVLN typology, active responses are either exit reactions that are destructive or voice reactions that are constructive. However, active attempts to change the situation, that is, voice, can take various forms (Saunders, Sheppard, Knight and Roth, 1992) di€ering in their degree of constructiveness. From the literature on con¯ict management it can be learned that actions to resolve a con¯ict which are comparable to voice responses, can be divided into two categories: problem solving and contending (Blake and Mouton, 1964; Rubin et al., 1994). Whereas problem solving entails e€orts to develop and move toward a solution that appeals to both sides, contending entails e€orts to resolve a con¯ict on one's own terms without 1 In

addition to passive loyalty, Withey and Cooper (1992) also distinguished active loyalty, which is similar to organizational citizenship behavior (Organ, 1988). However, active loyalty, including `Actively promote the organization's business in public' or `Do things above and beyond the call without being asked' does not seem a logical reaction to a problematic situation, and therefore is not examined in the present study.

Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

EMPLOYEES' REACTIONS TO PROBLEMATIC EVENTS

311

regard to the other party's interests. Problem solving tends to de-escalate a con¯ict, which is bene®cial to the organization, while contending has a tendency to escalate a con¯ict, which may be detrimental to the e€ectiveness of an organization (Rubin et al., 1994). Accordingly, voice should be divided into two forms: a constructive form, which consists of attempts to solve the problem considering one's own concerns as well as those of the organization (considerate voice), and a less constructive form, which consists of e€orts to win, without consideration for the concerns of the organization (aggressive voice). Examples of considerate voice are `Discussing the problem with the supervisor' and `Talking with your supervisor about the problem until you reach total agreement', and examples of aggressive voice are `Being persistent with your supervisor in order to get what you want' and `Starting a `®ght' with your supervisor'. In addition to this theoretical problem, a related methodological issue needs to be addressed, that is, the internal consistencies of the measures. Prior to studies (Rusbult et al., 1988; Withey and Cooper, 1989) reported low Cronbach's alphas for the measures of voice, patience and neglect (ranging from 0.41 to 0.77). In the present study, new scales were developed to examine the extent to which the ®ve categories of behavioral reactions are not only theoretically but also empirically separable. Furthermore, we explored whether these ®ve categories can be described along the two dimensions of the EVLN model Farrell (1983) proposed, assuming that aggressive voice is a form of trying to change the problematic situation that is less constructive than considerate voice, but more constructive than exit. Speci®cally, it was expected that reactions to problematic events can be depicted in a two-dimensional circular space, that is, a circumplex structure, formed by the coordinates of the destructive±constructive and active±passive dimensions. In other words, this circumplex structure is a conceptual representation of the domain of behavioral reactions that depicts behavioral variables as a combination of the two dimensions (cf. Kiesler, 1983; Wiggins, 1979). Prior studies have addressed the question of under what conditions employees respond with exit, voice, patience, or neglect (Leck and Saunders, 1992; Rusbult et al., 1988; Withey and Cooper, 1989). In these studies, di€erent predictor variables, such as investment size, alternatives for the job, locus of control and organizational commitment, were investigated. However, one of the most frequently examined predictor variables was job satisfaction. General job satisfaction was considered as a stimulator of constructive responses and as a suppressor of destructive responses. For example, Rusbult et al. (1988) reasoned that employees who are generally more satis®ed with their jobs probably feel strongly motivated to restore good working conditions and probably feel more optimistic about the possibilities of improving working conditions, and therefore are more likely to respond constructively than employees who are generally less satis®ed with their jobs. The results of the above-mentioned studies support the suppressor/stimulator role of job satisfaction with regard to voice, neglect and exit, but with regard to patience (passive loyalty) the results are inconclusive. Leck and Saunders (1992) and Farrell and Rusbult (1992) found positive links between prior satisfaction and patience, while Withey and Cooper (1989) found a negative link. Interview results of the study of Withey and Cooper (1989) suggested that this negative link could be explained by feelings of entrapment. Passive loyalists seemed to be entrapped persons who were just biding their time. Compatible with this ®nding, Withey and Coopers' (1989) measures did not re¯ect employees' faith in the organization and feelings of optimism. In the current study, congruent with our de®nition of patience, the operationalization of patience entails optimism and faith in the organization. Therefore, in line with Farrell and Rusbult (1992), we expected a positive relation between job satisfaction and patience. Moreover, the link between job satisfaction and considerate voice was also expected to be positive, while the relations between job satisfaction and exit, aggressive voice, and neglect were expected to be negative. Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

312

M. HAGEDOORN ET AL.

Figure 1. Five types of reactions of employees to problematic events, in terms of two descriptive dimensions

The present study examined the relationships between the ®ve dimensions of the Job Descriptive Index (Smith, Kendall and Hulin, 1969), that is, satisfaction with the work itself, supervision, payments, promotions, and co-workers, on the one hand, and the ®ve responses to a problematic event, on the other hand. The purpose of the investigation of these relations was twofold: (1) exploration of the stimulator/suppressor role of various aspects of job satisfaction, and (2) additional testing of the validity of the descriptive pattern of interrelations picture in Figure 1. The assumed circumplex structure of behavioral responses to a problematic event implies that associations with a predictor or dependent variable, such as job satisfaction, ®rst decrease from the most positively (or least negatively) associated behavioral response category and then increase again as one goes around the circular structure of behavioral response types (cf. Schwartz, 1992). In other words, the circumplex structure would be supported indirectly if with respect to all aspects of job satisfaction, the correlations with behavioral intentions follow a circular sequence.

Method Samples and procedure Data were collected from two groups: teachers and maternity nurses.

Teachers Teachers at a school for technical and vocational training for 16- to 18-year-olds were informed about the study by a newsletter and by their supervisors and the ®rst author of this article during sta€ meetings. They were told that the purpose of the study was to investigate on-the-job experiences of teachers and potential stressors at work and the results of the study could be used to indicate aspects of the work situation that needed to be improved. In total, 147 teachers ®lled out the questionnaire (response rate was 41 per cent). The sample consisted of 80 men and 66 women; one subject did not indicate his or her gender. The average age was 43 years (ranging Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

EMPLOYEES' REACTIONS TO PROBLEMATIC EVENTS

313

from 28 to 58 years) and the average length of employment was 12.3 years (ranging from 3 months to 32 years). On average, the subjects were employed for 29 hours a week. The response rate was rather low, but the sample of teachers is representative of the population with regard to gender, age, and the number of hours the teachers were employed. The percentage of men and women in the sample and the population are identical (men: 54 per cent; women 46 per cent). In the population, the mean age was 42.5 years and the average number of hours the teachers were employed was 28.5 hours a week. Maternity nurses2 Questionnaires were mailed to 106 maternity nurses after they were informed about the purpose of the study by a newsletter. The task of these maternity nurses consisted of taking care of a mother and her newly-born child in the home and doing some housework. A total of 87 female maternity nurses ®lled out the questionnaire (response rate: 82 per cent). The average age of the subjects was 35 years (ranging from 22 to 53 years) and the average length of employment was 8 years (ranging from 1 year to 27 years). On average, participants were employed for 30 hours a week.

Measures Job satisfaction Job satisfaction was measured by the Job Descriptive Index developed by Smith et al. (1969). This measure consists of ®ve subscales that assess di€erent aspects of job satisfaction, namely, the work itself, supervision, pay, promotions, and co-workers. The respondents completed the items with 3-point scales (yes, no, or ?). Agreement responses to positive items and disagreement responses to negative items received a score of 3; disagreement responses to positive items and agreement responses to negative items received a score of 0; and the ? received a score of 1 (Smith et al., 1969). The ? is not scored on the middle of the scale because Smith et al. (1969) found that individuals classi®ed as dissatis®ed answered the items with the ? more often than individuals classi®ed as satis®ed. Recently, Hanisch (1992) demonstrated with Item Response Theory analyses that this scoring procedure is still justi®ed today. The JDI subscales had acceptable levels of internal consistencies (see Table 3).3 The ®ve categories of responses to problematic events The di€erent responses were introduced as follows. `Everybody occasionally encounters a problem or a problematic event at work. This can be a di€erence of opinion with your supervisor, frustrations with regard to the behavior of co-workers, or dissatisfaction, for instance, about a schedule or a speci®c task you are assigned to do. People tend to react di€erently to these experiences. On the following pages, several descriptions of possible reactions are listed. Would you indicate how likely it is that you would react to problematic events in the described ways?' The items were completed with 7-point scales (1 ˆ de®nitely not, to 7 ˆ de®nitely yes). Items that 2 The

authors are grateful to Marco Bonarius, Rigobert Bromiscu, Rixt de Lang, and Saapke Postma for their help in data collection. 3 The Cronbach's alphas for the two subsamples separately were approximately the same, except for satisfaction with promotion, which was 0.78 for the subsample of teachers and 0.17 for the subsample of maternity nurses.

Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

314

M. HAGEDOORN ET AL.

Table 1. Oblique rotation factor analysis of items to measure responses to a problematic event Items Exit 1. Consider possibilities to change jobs 2. Actively look for a job outside the ®eld of education/health care 3. Intend to change employers 4. Actively look for a job elsewhere within the ®eld of education/health care 5. Look for job advertisements in newspapers to which you could apply 6. Intend to change your ®eld of work Considerate voice 7. Try to come to an understanding with your supervisor 8. In collaboration with your supervisor, try to ®nd a solution that is satisfactory to everybody 9. Try to work out an ideal solution in collaboration with your supervisor 10. Together with your supervisor, explore each other's opinions until the problems are solved 11. Try to compromise with your supervisor 12. Talk with your supervisor about the problem until you reach total agreement 13. Suggest solutions to your supervisor 14. Immediately report the problem to your supervisor 15. Immediately try to ®nd a solution 16. Try to think of di€erent solutions to the problem 17. Ask your supervisor for a compromise Patience 18. Trust the decision-making process of the organization without your interference 19. Trust the organization to solve the problem without your help 20. Have faith that something like this will be taken care of by the organization without you contributing to the problem-solving process 21. Assume that in the end everything will work out 22. Optimistically wait for better times Aggressive voice 23. Describe the problem as negatively as possible to your supervisor 24. Try to win the case 25. Deliberately make the problem sound more problematic than it really is 26. Being persistent with your supervisor in order to get what you want

Factor 1

Factor 2

Factor 3

Factor 4

Factor 5

0.863 0.833

ÿ0.004 0.026

ÿ0.033 ÿ0.046

ÿ0.037 ÿ0.048

0.105 0.070

0.796 0.793

ÿ0.065 0.021

ÿ0.072 0.166

0.092 0.025

0.121 ÿ0.102

0.776

0.003

ÿ0.121

0.022

0.173

0.724

ÿ0.120

ÿ0.075

0.167

0.161

ÿ0.099

0.807

0.126

0.037

0.020

0.084

0.787

0.060

0.098

ÿ0.150

0.086

0.779

0.087

ÿ0.095

ÿ0.113

0.110

0.753

0.082

ÿ0.046

ÿ0.064

ÿ0.020 ÿ0.082

0.738 0.731

ÿ0.089 0.229

ÿ0.026 0.063

ÿ0.085 ÿ0.112

ÿ0.039 ÿ0.059

0.681 0.591

ÿ0.183 ÿ0.018

0.020 ÿ0.092

0.093 0.259

ÿ0.128 ÿ0.112

0.536 0.500

ÿ0.119 ÿ0.221

ÿ0.067 0.018

0.052 0.128

0.163

0.480

ÿ0.035

0.200

ÿ0.148

0.040

0.133

0.751

ÿ0.071

ÿ0.213

ÿ0.006

ÿ0.119

0.702

ÿ0.052

0.174

0.050

0.028

0.702

ÿ0.168

ÿ0.014

ÿ0.008

ÿ0.043

0.648

0.073

0.056

ÿ0.129

0.005

0.492

0.111

0.096

ÿ0.052

ÿ0.141

0.013

0.808

0.066

0.015 0.113

0.063 ÿ0.084

ÿ0.155 ÿ0.018

0.730 0.697

ÿ0.010 0.173

0.052

0.222

0.129

0.684

0.056

Table continues on next page Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

EMPLOYEES' REACTIONS TO PROBLEMATIC EVENTS

315

Table 1. Continued Items 27. Starting a `®ght' with your supervisor 28. Try to prove in all possible ways to your supervisor that you are right 29. By de®nition, blame the organization for the problem Neglect 30. Report sick because you do not feel like working 31. Come in late because you do not feel like working 32. Put less e€ort into your work than may be expected of you 33. Now and then, do not put enough e€ort into your work 34. Missing out on meetings because you do not feel like attending them Eigenvalues Percentage explained variance

Factor 1

Factor 2

Factor 3

Factor 4

Factor 5

0.152 ÿ0.047

0.115 0.095

ÿ0.231 0.019

0.615 0.602

ÿ0.123 0.190

0.074

ÿ0.206

0.116

0.595

ÿ0.077

0.246

0.046

ÿ0.005

ÿ0.031

0.743

0.093

0.091

0.115

ÿ0.023

0.635

0.089

ÿ0.161

ÿ0.008

0.119

0.632

0.211

ÿ0.046

0.016

0.190

0.621

0.001

ÿ0.137

0.034

0.278

0.548

7.653 22.5

5.174 15.2

2.489 7.3

2.177 6.4

1.624 4.8

assessed the di€erent categories of responses were based on Farrell (1983), Rusbult et al. (1988), and Van de Vliert (1997). Table 1 shows examples of these items. Exit was measured by seven items. Considerate voice was assessed by 12 items, including eight items based on the subscales problem solving and compromising of the DUTCH (Dutch Test of Con¯ict Handling; Van de Vliert, 1997). For the items based on the DUTCH, see Table 1 item numbers 7 to 12 and 17. Aggressive voice was measured by 11 items of which four were based on the subscale, contending, of the DUTCH (Table 1, item numbers 24, 26, 27, 28). Patience was measured by six items, and neglect was assessed by 11 items.

Results In order to investigate whether the ®ve responses to a problematic event are not only theoretically, but also empirically separable, the items were factor analyzed using a principal components method. Prior to this factor analysis, by means of a Simultaneous Components Analysis (SCA; Millsap and Meridith, 1988; Kiers and Ten Berge, 1989), it was examined whether the subsample of teachers and the subsample of maternity nurses could be analyzed as one group. In SCA, component weights are found that de®ne identical components which optimally account for the variance in both samples simultaneously. It is justi®ed to treat two groups as one sample when for each group the variances explained by the SCA solution are approximately the same as the variances explained by the principal component solution for that group. The SCA showed that compared to separate principal components analyses the explained variances decreased with only 0.68 per cent (from 50.14 per cent to 49.46 per cent) for the subsample of teachers and with 0.61 per cent (from 69.19 per cent to 68.58 per cent) for the subsample of maternity nurses. Thus, it is justi®ed to factor analyze the two subsamples as one group. The subject-to-item ratio in the Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

316

M. HAGEDOORN ET AL.

subject factor analysis was an acceptable 5 : 1 (Gorsuch, 1983; cf. Ford, MacCallum and Tait, 1986). Because of the theoretical assumptions and a scree test, a 5-factor solution with factors allowed to be correlated was speci®ed (oblique rotation; cf. Ford et al., 1986). This factor solution explained 48.8 per cent of the variance. Items de®ning a factor were those with structure coecients greater than 0.45. In line with the theoretical assumptions, the ®ve factors could be labeled exit, considerate voice, patience, aggressive voice, and neglect. Eleven of the 47 items had factor loadings smaller than 0.45 or cross-loaded (one exit item, one considerate voice item, four aggressive voice items, one patience item and six neglect items), and were therefore dropped. Furthermore, in order to improve face validity, two items (`Show your irritation to the supervisor' and `Make demands to your supervisor') were dropped because they were meant to indicate aggressive voice, but loaded on the considerate voice factor. After excluding these 13 items, a second factor analysis with the remaining 34 items was executed. Again, ®ve factors were speci®ed that now explained 56.2 per cent of the variance. Table 1 shows the items, factor loadings, eigenvalues, and the percentage of variance explained by each factor. Table 2 shows the factor correlations. Table 2. Interfactor collections

1. 2. 3. 4. 5.

Factor Factor Factor Factor Factor

1 2 3 4 5

2

3

4

5

ÿ0.096 Ð

ÿ0.052 ÿ0.106 Ð

0.355 ÿ0.016 ÿ0.124 Ð

0.239 ÿ0.124 ÿ0.003 ÿ0.257 Ð

In contrast to prior studies (e.g. Rusbult et al., 1988; Withey and Cooper, 1989), the Cronbach's alphas of the scales for responses to a problematic event used in the current study were high (50.80). Cronbach's alpha for the patience scale was lower (0.69), and can still be improved. Table 3 shows the reliabilities, means, standard deviations, and numbers of items of all the dependent as well as the independent variables for the current sample.4 On average, the respondents indicated that they were most likely to respond to a problematic event with considerate voice and were least likely to respond with neglect. Furthermore, the mean satisfaction score was highest with respect to the work itself and lowest with regard to promotion. In order to investigate whether the ®ve behavioral responses to a problematic event can be described along the two dimensions described by Farrell (1983), a second-order factor analysis (i.e. factor analysis on the scale scores) with factors allowed to be correlated was performed (oblique rotation; cf. Ford et al., 1986). Two factors were speci®ed, which together explained 62 per cent of the variance. The eigenvalues were 2.00 and 1.11 for factor 1 and factor 2, respectively. Indeed, in line with the empirical ®ndings of Farrell (1983), the dimensions can be interpreted as destructive±constructive (factor 1) and active±passive (factor 2). Figure 2 shows

4 These

descriptive statistics were also computed for the two subsamples separately. There were some di€erences with respect to the means. The results of a Multivariate Variance Analysis showed that teachers were more likely to use aggressive voice and neglect than maternity nurses. Furthermore, teachers were less satis®ed with the work itself, supervision, pay and co-workers than maternity nurses (F(10,193) ˆ 11.913; p 5 0.001). These di€erences were not due to gender; there were no di€erences between the means of all the variables examined in this study for male and female teachers (F(10,116) ˆ 1.707; n.s.). Despite the di€erences in mean scores, the data of the subsamples show identical patterns. The internal consistencies of the scales for the ®ve responses were approximately the same.

Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

EMPLOYEES' REACTIONS TO PROBLEMATIC EVENTS

317

Table 3. Descriptive statistics for all dependent and independent variables Variables Considerate voice Aggressive voice Exit Neglect Patience Satisfaction with work Satisfaction with supervision Satisfaction with pay Satisfaction with promotion Satisfaction with co-workers

N of items

Mean

S.D.

Cronbach's alpha

11 7 6 5 5 18 18 9 9 18

5.01 2.89 2.90 1.92 3.50 2.19 1.88 1.88 0.74 2.07

0.87 0.97 1.39 0.88 0.98 0.54 0.75 0.68 0.47 0.64

0.88 0.83 0.92 0.79 0.69 0.83 0.91 0.81 0.65 0.90

that the two-dimensional plot of the ®ve responses that results from the second-order factor analysis is not a perfect replication of Figure 1, but both patterns seem similar. Although exit and aggressive voice are less active than expected and neglect is less passive than expected, Figure 2 supports the presumed circumplex structure of responses to problematic events. Lastly, the correlations between all variables were computed (Table 4). In line with Figure 2, signi®cant correlations are observed between exit, aggressive voice, and neglect (ranging from 0.45 to 0.49). Of particular importance are the correlations between the di€erent aspects of job satisfaction, on the one hand, and employees' behavioral reactions, on the other hand. In line with our expectations, all aspects of satisfaction, with the exception of pay, are linked positively to considerate voice and patience, and negatively to exit, aggressive voice and neglect. This is especially true for satisfaction with supervision. It was expected that the ®ve responses could be ordered in a circumplex structure running from considerate voice via aggressive voice to exit, and back to considerate voice via neglect and patience (see Figure 1). On the basis of this structure, the correlations with other variables, such as job satisfaction, should follow a circular path. Because the pattern of intercorrelations is tested

Figure 2. Two-dimensional picture of the ®ve responses to a problematic event as a result of a second-order factor analysis Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

318

M. HAGEDOORN ET AL.

Table 4. Zero-order correlations between the ®ve categories of responses and the ®ve aspects of job satisfaction Variables 1. 2. 3. 4. 5. 6. 7.

Considerate voice Aggressive voice Exit Neglect Patience Satisfaction with work Satisfaction with supervision 8. Satisfaction with pay 9. Satisfaction with promotion 10. Satisfaction with co-workers

2 ÿ0.01 Ð

3

4

5

ÿ0.13* ÿ0.21{ ÿ0.07 0.45{ 0.47{ ÿ0.12 Ð 0.49{ ÿ0.07 Ð ÿ0.01 Ð

6

7

8

9

10

0.17{ 0.43{ 0.12 0.28{ 0.11 ÿ0.30{ ÿ0.42{ ÿ0.04 ÿ0.21{ ÿ0.39{ ÿ0.35{ ÿ0.44{ ÿ0.21{ ÿ0.33{ ÿ0.31{ ÿ0.27{ ÿ0.35{ ÿ0.01 ÿ0.17* ÿ0.29{ 0.13* 0.26{ 0.07 0.08 0.23{ Ð 0.43{ 0.06 0.33{ 0.33{ Ð 0.14* 0.36{ 0.47{ Ð

0.26{ Ð

0.11 0.13* Ð

* p 5 0.05; { p 5 0.01; { p 5 0.001.

and not the statistical signi®cance of single correlations, even nonsigni®cant correlations provide meaningful information. The correlations with satisfaction with work, for example, did follow a circular path from 0.17 (considerate voice) via ÿ0.30 (aggressive voice) to ÿ0.35 (exit), and via ÿ0.27 (neglect) and 0.13 (patience) back to 0.17. In order to test the overall circular pattern with respect to job satisfaction, a Friedman test (Siegel, 1956) was employed. With respect to each scale of job satisfaction, the ®ve behavioral responses were ranked on the basis of their correlation with job satisfaction (i.e. from most positively related to job satisfaction to most negatively related to job satisfaction). The order of the reactions was more or less identical for the di€erent aspects of job satisfaction (see also Table 4). Next, for each behavioral reaction the rank scores were summed. The sums of the ranks for the ®ve behavioral responses di€er signi®cantly (wr2 ˆ 18.72, p 5 0.001).5 Thus, the circular pattern of correlations proves to be signi®cant. Put di€erently, the Friedman test provides additional support for the circumplex structure of the reactions to a problematic event examined here.

Discussion In the present study, the EVLN typology of Farrell (1983), re®ned on the basis of the literature on con¯ict management, was tested. First of all, the results support the idea that the category of voice responses de®ned by Farrell (1983) should be divided into two forms. One form is considerate voice that consists of actions that take the concerns of the actor as well as the concerns of the organization into account (cf. problem solving, Rubin et al., 1994). The second form, aggressive voice, only represents the one-sided considerations of the employee (cf. contending, Rubin et al., 1994). In contrast to previous research (Rusbult et al., 1988; Withey and 5 Although

there were some di€erences in absolute strength of intercorrelations in the two subsamples, identical patterns were observed. The Friedman test was also performed for both subsamples separately; wr2 was 18.40 ( p 5 0.01) for the teachers and wr2 was 13.60 ( p 5 0.01) for the maternity nurses.

Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

EMPLOYEES' REACTIONS TO PROBLEMATIC EVENTS

319

Cooper, 1989), all the subscales of behavioral responses, that is, exit, considerate voice, aggressive voice, patience, and neglect, are reliable in terms of internal consistency, and they are not only theoretically but also empirically separable from each other. However, the internal consistency of the patience scale (0.69) can still be improved. Furthermore, it has also been shown that the ®ve behavioral responses form a circumplex structure. More speci®cally, they can be depicted in a circular space as coordinates of the dimensions destructive±constructive and active±passive as proposed by Farrell (1983; cf. Rusbult et al., 1988).6 In line with prior studies (Leck and Saunders, 1992; Rusbult et al., 1988; Withey and Cooper, 1989), job satisfaction seems to promote constructive responses (considerate voice and patience) and suppress destructive responses (exit, aggressive voice and neglect) to a problematic event. With regard to patience, these results seem to contradict the ®ndings of Withey and Cooper (1989). However, as mentioned earlier, their operationalization of passive loyalty seems to re¯ect feelings of entrapment. In contrast, in the present study, patience was measured with items re¯ecting optimism and faith in the organization. Thus, patience de®ned as optimistically waiting until conditions improve, trusting the organization to do the right thing seems to be promoted by prior satisfaction. This leads to the conclusion that comparable to the voice category, it might also be useful to divide passive loyalty into two forms, namely a constructive and a destructive form (Withey and Cooper, 1992). The circumplex structure of behavioral responses to a problematic event predicts that the destructive form can be placed between patience (i.e. the constructive form of passive loyalty) and neglect. Corresponding to our expectations, the two distinct forms of voice were related di€erently to job satisfaction; aggressive voice was linked negatively to satisfaction and considerate voice was linked positively to job satisfaction. Thus, employees who are less satis®ed as well as those who are more satis®ed make use of voice, but they appear to prefer di€erent forms. This ®nding provides further support for the re®nement of the EVLN model. Considering the positive relationship between voice and satisfaction found in previous studies, the results of the present study suggest that in those studies, voice measures re¯ected considerate voice. Apparently, the voice category of Farrell (1983; Hirschman, 1970) is too heterogeneous and should be divided into at least two separate categories, namely, considerate voice and aggressive voice. In this study, these two forms were operationalized as forms of reporting problems to authorities within the organization (i.e. internal report of problems). In addition to internal report, forms of external report can be identi®ed, such as whistle-blowing (cf. Van Dyne, Cummings and Mclean Parks, 1995). This external report may be with or without regard for the interests of the organization, which means that external reporters may intend to gain rewards for themselves or may intend to bene®t the organization. Therefore, external report may resemble considerate voice or aggressive voice, and accordingly can be placed in the circumplex structure nearby considerate or aggressive voice. Concerning the promotor/suppressor role of job satisfaction, especially satisfaction with supervision shows strong positive links with considerate voice and patience and strong negative 6 The

label `destructive±constructive' is not purely descriptive but also refers to the outcomes of the responses, that is, the impact on the employee±organization relationship and the problem at hand, while this is not true for the label `active± passive'. Behaviors described in terms of outcomes should not be used to predict those same outcomes; it confounds independent and dependent variables. Consequently, researchers who want to study the outcomes of the ®ve responses need purely descriptive dimensions. Farrell's (1983) active±passive dimension meets this condition (cf. Van de Vliert and Euwema, 1994). Van de Vliert and Euwema (1994) showed that agreeableness is an appropriate second dimension to describe con¯ict behaviors that are similar to the behavioral responses to a problematic event. Agreeableness refers to the extent to which responses are pleasant and relaxed rather than unpleasant and strained. Thus, one might consider using the label `disagreeable±agreeable' instead of the label `destructive±constructive'.

Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

320

M. HAGEDOORN ET AL.

links with exit, aggressive voice, and neglect. One explanation for these ®ndings might be that con®dence in possibilities for improvement is increased more by satisfaction with supervision than by satisfaction with work, pay, promotions, and co-workers because most changes have to be permitted or implemented by supervisors. Consistent with the argument that the behavior of the supervisor plays an important role with regard to employees' reactions to a problematic event, Saunders and his colleagues (1992) showed that the employees who perceived their supervisor as more approachable and responsive were more likely to voice their concerns. Similarly, Parker (1993) demonstrated a positive link between the perception that supervisors take the concerns of employees into account and the willingness to use reformist dissent (cf. considerate voice), while they found a negative link between this perception and exit. The perception of responsiveness and approachability of supervisors re¯ects a sense of justice (Saunders et al., 1992). This suggests that it is not the a€ective aspect of satisfaction, but the cognitive aspect, namely, the perception of justice, particularly procedural justice, that is responsible for the strong link between satisfaction and responses to a problematic event (cf. Moorman, 1991; Organ, 1988). Therefore, in future research, the role of procedural justice as a motivator of constructive responses and as a suppressor of destructive responses to a problematic even should be investigated more explicitly. To conclude, the present study showed that the ®ve examined behavioral responses to a problematic event form a circumplex structure on the basis of which additional categories can be identi®ed, including the four neglect categories of Robinson and Bennett (1995; cf. Fischer and Locke, 1992).7 We concentrated on the voice category and re®ned the EVLN model (Farrell, 1983) by di€erentiating considerate voice from aggressive voice. In addition, the comparison of our results to those of Withey and Cooper (1989, 1992) indicated the usefulness of a distinction between a constructive and destructive form of passive loyalty.

References Blake, R. R. and Mouton, J. S. (1964). The Managerial Grid, Gulf, Houston, TX. Farrell, D. (1983). `Exit, voice, loyalty, and neglect as responses to job dissatisfaction: A multidimensional scaling study', Academy of Management Journal, 26, 596±607. Farrell, D. and Rusbult, C. E. (1992). `Exploring the exit, voice, loyalty, and neglect typology: The in¯uence of job satisfaction, quality of alternatives, and investment size', Employee Responsibilities and Rights Journal, 5, 201±218. Fischer, C. D. and Locke, E. A. (1992). `The new look in job satisfaction research and theory'. In: Cranny, C. J., Smith, P. C. and Stone, E. F., (Eds) Job Satisfaction: How People Feel About Their Jobs and How It A€ects Their Performance, Lexington Books, New York, pp. 165±194. Ford, J. K., MacCallum, R. C. and Tait, M. (1986). `The application of exploratory factor analysis in applied psychology: A critical review and analysis', Personnel Psychology, 39, 291±314. Gorsuch, R. L. (1983). Factor Analysis, Lawrence Earlbaum Associates, Hillsdale, NJ. Hanisch, K. A. (1992). `The job descriptive index revisited: Questions about the question mark', Journal of Applied Psychology, 77, 377±382. Hirschman, A. O. (1970). Exit, Voice, and Loyalty: Responses to Decline in Firms, Organizations, and States, Harvard University Press, Cambridge, MA.

7 Robinson

and Bennett (1995) developed a typology of four categories of deviant workplace behaviors, namely, production deviance, property deviance, political deviance, and personal aggression. The ®nal neglect scale in the present study is similar to production deviance, while the items excluded on the basis of factor analyses seem to re¯ect property deviance (e.g. the item `Take school property home for your own use') and political deviance (e.g. `Deliberately avoid your supervisor' and `Vent o€ your tension on colleagues, pupils, or supervisors').

Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)

EMPLOYEES' REACTIONS TO PROBLEMATIC EVENTS

321

Hulin, C. (1991). `Adaptation, persistence, and commitment in organizations'. In: Dunnette, M. D. and Hough, L. M. (Eds) Handbook of Industrial and Organizational Psychology, Vol. 2, Consulting Psychologists Press, Palo Alto, pp. 445±505. Kahn, R. L. and Byosiere, P. (1992). `Stress in organizations'. In: Dunnette, M. D. and Hough, L. M., (Eds) Handbook of Industrial and Organizational Psychology, Vol. 3, Consulting Psychologists Press, Palo Alto, pp. 571±650. Kiers, H. A. L. and Ten Berge, J. M. F. (1989). `Alternating least squares algorithms for simultaneous components analysis with equal component weight matrices in two or more populations', Psychometrika, 54, 467±473. Kiesler, D. J. (1983). `The 1982 interpersonal circle: a taxonomy for complementarity in human transitions', Psychological Review, 90, 185±214. Leck, J. D. and Saunders, D. M. (1972). `Hirschman's loyalty: Attitude or behavior?' Employee Responsibilities and Rights Journal. 5, 219±230. Millsap, R. E. and Meridith, W. (1988). `Component analysis in cross-sectional and longitudinal data', Psychometrika, 53, 123±134. Moorman, R. H. (1991). `Relationship between organizational justice and organizational citizenship behaviors: Do fairness perceptions in¯uence employee citizenship?' Journal of Applied Psychology, 76, 845±855. Organ, D. W. (1988). `A restatement of the satisfaction±performance hypothesis', Journal of Management, 14, 547±557. Parker, E. L. (1993). `When to ®x it and when to leave: Relationships among perceived control, self-ecacy, dissent, and exit', Journal of Applied Psychology, 78, 949±959. Robinson, S. L. and Bennett, R. J. (1995). `A typology of deviant workplace behaviors: A multidimensional scaling study', The Academy of Management Journal, 38, 555±572. Rubin, J. Z., Pruitt, D. G. and Kim, S. H. (1994). Social Con¯ict: Escalation, Stalemate, and Settlement, McGraw-Hill, New York. Rusbult, C. E. and Lowery, D. (1985). `When bureaucrats get the blues: Responses to dissatisfaction among federal employees', Journal of Applied Social Psychology, 68, 429±438. Rusbult, C. E., Farrell, D., Rogers, G. and Mainous III, A. G. (1988). `Impact of exchange variables on exit, voice, loyalty, and neglect: An integrative model of responses to declining job satisfaction', Academy of Management Journal, 31, 599±627. Saunders, D. M., Sheppard, B. H., Knight, V. and Roth, J. (1992). `Employee voice to supervisors', Employee Responsibilities and Rights Journal, 5, 241±259. Schwartz, S. H. (1992). `Universals in the content and structure of values: Theoretical advances and empirical tests in 20 countries'. In: Zanna, M., (Ed.) Advances in Experimental Social Psychology, Vol. 25, Academic Press, San Diego, pp. 1±65. Sheppard, B. H., Lewicki, R. J. and Minton, J. W. (1992). Organizational Justice: The Search for Fairness in the Workplace, Lexington Books, New York. Siegel, S. (1956). Nonparametric Statistics for the Behavioral Sciences, McGraw-Hill Kogakusha, Tokyo. Smith, P. C., Kendall, L. M. and Hulin, C. L. (1969). The Measurement of Satisfaction in Work and Retirement, Rand McNally, Chicago. Van de Vliert, E. (1997). Complex Interpersonal Con¯ict Behavior: Theoretical Frontiers, Psychology Press, Hove. Van de Vliert, E. and Euwema, M. C. (1994). `Agreeableness and activeness as components of con¯ict behaviors', Journal of Personality and Social Psychology, 66, 674±687. Van Dyne, L., Cummings, L. L. and Mclean Parks, J. (1995). `Extra-role behaviors: In pursuit of construct and de®nitional clarity (a bridge over muddied waters)'. In: Cummings, L. L. and Staw, B. M. (Eds) Research in Organizational Behavior, Vol. 17, JAI Press, Greenwich, CT, pp. 215±285. Wiggins, J. S. (1979). `A psychological taxonomy of trait-descriptive terms: The interpersonal domain', Journal of Personality and Society Psychology, 37, 395±412. Withey, M. J. and Cooper, W. H. (1989). `Predicting exit, voice, loyalty, and neglect', Administrative Science Quarterly, 34, 521±539. Withey, M. J. and Cooper, W. H. (1992). `What's loyalty?' Employee Responsibilities and Rights Journal, 5, 231±240.

Copyright # 1999 John Wiley & Sons, Ltd.

J. Organiz. Behav. 20, 309±321 (1999)